Author

Topic: Hot cables on S3, what am I doing wrong? (Read 1905 times)

full member
Activity: 169
Merit: 100
July 30, 2015, 04:57:36 AM
#28
this is why bitmain should have used 8 pin PCI-E connectors instead of 6 pin.

the s3 is just at the edge of 200w per board, if you have a psu with 2 ends per cord, it will be overloaded.

its important to use a PSU that has 4 separate cables per s3.


ur right.
since my psu only have 2 pci-e, i change cpu connector (8 pin 12V) into pci-e.
its not easy to take pins out from connector, do it carefully.
there is another 12V pin at the atx connector, but its too complicated because the cable tied up as one.
drawback for this methode is unable to use psu for cpu directly.you have to reverse it before plug it to cpu.
simple way is use 4 cable psu(not 2 cable 4 connector).

S3+ as long as you are not overclocking a lot 2 cables should power it.  One pcie power cable to each side.

S5 is a different story 4 power cables is needed.  But it uses more watts.

A big thing is either one dont use a cheap PSU.  If you use a cheap psu it's cords are probley not enough guage and it will get hot.

that because i use "cheap" psu. it comes with 2 cable but 4 connector pcie(2x2 parallel)
first attempt cable get hot, after modification slightly warm.
sure cable gauge is the problem, psu component itself can handle the wattage(mine 650).
from april till now 3+ months work fine with 0 problem.

edit:not overclock. had oc before but it burned my psu(same psu with the last one).
legendary
Activity: 1456
Merit: 1000
this is why bitmain should have used 8 pin PCI-E connectors instead of 6 pin.

the s3 is just at the edge of 200w per board, if you have a psu with 2 ends per cord, it will be overloaded.

its important to use a PSU that has 4 separate cables per s3.


ur right.
since my psu only have 2 pci-e, i change cpu connector (8 pin 12V) into pci-e.
its not easy to take pins out from connector, do it carefully.
there is another 12V pin at the atx connector, but its too complicated because the cable tied up as one.
drawback for this methode is unable to use psu for cpu directly.you have to reverse it before plug it to cpu.
simple way is use 4 cable psu(not 2 cable 4 connector).

S3+ as long as you are not overclocking a lot 2 cables should power it.  One pcie power cable to each side.

S5 is a different story 4 power cables is needed.  But it uses more watts.

A big thing is either one dont use a cheap PSU.  If you use a cheap psu it's cords are probley not enough guage and it will get hot.
full member
Activity: 169
Merit: 100
this is why bitmain should have used 8 pin PCI-E connectors instead of 6 pin.

the s3 is just at the edge of 200w per board, if you have a psu with 2 ends per cord, it will be overloaded.

its important to use a PSU that has 4 separate cables per s3.


ur right.
since my psu only have 2 pci-e, i change cpu connector (8 pin 12V) into pci-e.
its not easy to take pins out from connector, do it carefully.
there is another 12V pin at the atx connector, but its too complicated because the cable tied up as one.
drawback for this methode is unable to use psu for cpu directly.you have to reverse it before plug it to cpu.
simple way is use 4 cable psu(not 2 cable 4 connector).
legendary
Activity: 1736
Merit: 1006
this is why bitmain should have used 8 pin PCI-E connectors instead of 6 pin.

the s3 is just at the edge of 200w per board, if you have a psu with 2 ends per cord, it will be overloaded.

its important to use a PSU that has 4 separate cables per s3.
legendary
Activity: 1302
Merit: 1068
One thing to keep in mind about most surge protectors - every time they protect vs a surge, it degrades them. Eventually they degrade to the point they are providing ZERO protection.

 Definitely applies to any surge protector using a MOV (or combo types with a MOV as the last stage, though those type last longer as the other circuitry soaks some of each surge leaving less applied to the MOV to degrade it).

 I remember there being at least one type that does NOT use a MOV and does NOT degrade, but I also remember that type being VERY EXPEN$$$$$IVE.


The one i'm using has leds and self test unit, i assume it will tell me if its not protecting since the led for it are there.

And i never had any surge trip the surge protection which force you to manually reset the power bar.

Do micro-surges damage the "MOV" too?
legendary
Activity: 1498
Merit: 1030
One thing to keep in mind about most surge protectors - every time they protect vs a surge, it degrades them. Eventually they degrade to the point they are providing ZERO protection.

 Definitely applies to any surge protector using a MOV (or combo types with a MOV as the last stage, though those type last longer as the other circuitry soaks some of each surge leaving less applied to the MOV to degrade it).

 I remember there being at least one type that does NOT use a MOV and does NOT degrade, but I also remember that type being VERY EXPEN$$$$$IVE.
legendary
Activity: 1302
Merit: 1068

I still wonder how come only one side fried and not everything.

We lose single ports in big rack mounted, ups protected switches all the time.  Same deal, path of least resistance.  It is weird.

Oh i see, somehow i thought it would be easier for lightning to propagate evenly instead of picking a path and shoving it through to the end.

Well i guess its better than losing all your ports on your whole rack. :S
full member
Activity: 198
Merit: 102

I still wonder how come only one side fried and not everything.

We lose single ports in big rack mounted, ups protected switches all the time.  Same deal, path of least resistance.  It is weird.
legendary
Activity: 1302
Merit: 1068
Hmm. Weird, the rating on this bar is enough for intense storm and it guarantee anything its connected to if it get fried. Still good to know, thanks.

I guess this is pretty much settled for you anyways, since you sold it.

I still wonder how come only one side fried and not everything.
full member
Activity: 198
Merit: 102
What firmware do you install in the S3?.

I had the same problem, and one January firmware have a issue with voltage. My S3 was burn and loss it.

I recommend install the last firmware if you have not OC or November firmware and update the CGMINER of Kano.

Regards.
Antuam

Well I never checked firm ware version or for updates.  I guess I will take a look at that, thx.
full member
Activity: 198
Merit: 102
How did you have the power plugged in?  Did you have 2 pcie plugs plugged in from same power cord on each side?   IE one power cord and split into 2 pcie plugs.
The corsair has 4 individual cables with pcie connectors coming out of it, i have two in each miner.  I don't know about the rails on the inside of the PSU though.  They may be basically together inside the case.  Im going to switch to those Dells as soon as I can get some PCIe cables in.

I am thinking that there was a voltage surge and your PSU got damaged and let too much current into the miner. Generally its not a good idea to keep miners running during a storm.
Yes I agree that is probably what happened to the one with the board that quit working.  Its in the server room at my office.  Living in FL, I can't unplug during every storm or it would never be plugged in LOL.

Since all the heat seems to be concentrated at the connection you should of had a bad connection on the pin or corrosion that caused them to over heat.  S3+ should not have problems like that even with 18 AWG.
I pay pretty close attention to detail and I didn't see anything out of the ordinary on the connector or the pins.  I also thought an s3+ shouldn't have problems like that on an 850 watt psu, granted there we're 2 s3s on this psu but still.

I see, this must be the miner we discussed. I would like to know, were the PSU plugged in a surge protection power bar? I have mine on and i assumed they are safe from surges.

It was in a cheap power bar but I find here in Fl where the Lightning is literally an every day occurrence, nothing can stop that kind of voltage.  These are in our server room at my office and the whole room is surge protected as well by a big surge protector that built in the wall by the power distribution panel.  As well as another whole building surge protector that is in a utility room.  We lose equipment every summer even with all that.  We have never had a surge get to the racks which are behind big and rAck mount ups though.
legendary
Activity: 1302
Merit: 1068
How did you have the power plugged in?  Did you have 2 pcie plugs plugged in from same power cord on each side?   IE one power cord and split into 2 pcie plugs.
The corsair has 4 individual cables with pcie connectors coming out of it, i have two in each miner.  I don't know about the rails on the inside of the PSU though.  They may be basically together inside the case.  Im going to switch to those Dells as soon as I can get some PCIe cables in.

I am thinking that there was a voltage surge and your PSU got damaged and let too much current into the miner. Generally its not a good idea to keep miners running during a storm.
Yes I agree that is probably what happened to the one with the board that quit working.  Its in the server room at my office.  Living in FL, I can't unplug during every storm or it would never be plugged in LOL.

Since all the heat seems to be concentrated at the connection you should of had a bad connection on the pin or corrosion that caused them to over heat.  S3+ should not have problems like that even with 18 AWG.
I pay pretty close attention to detail and I didn't see anything out of the ordinary on the connector or the pins.  I also thought an s3+ shouldn't have problems like that on an 850 watt psu, granted there we're 2 s3s on this psu but still.

I see, this must be the miner we discussed. I would like to know, were the PSU plugged in a surge protection power bar? I have mine on and i assumed they are safe from surges.
legendary
Activity: 1722
Merit: 1000
wall outlet > ups > power supply

UPS will eat the power spikes.
legendary
Activity: 1722
Merit: 1005
What firmware do you install in the S3?.

I had the same problem, and one January firmware have a issue with voltage. My S3 was burn and loss it.

I recommend install the last firmware if you have not OC or November firmware and update the CGMINER of Kano.

Regards.
Antuam
sr. member
Activity: 362
Merit: 250
I use the CX500 on my s3's. It's way cheaper to buy many of them. I ran my s3's at 250 until recently went back to stock to accommodate summer heat. Cables would only get mildly warm. You can get a CX500 for $25 currently after $20 rebate.
The CX series is a lower tier psu then the one that OP is using, this isn't a good idea or at least if you home mine and don't want anything to catch fire while you're asleep. I say test the miner and make sure you are using all four pci plugs, weird this happened but I suspect there is an issue on the upgrade kit, could be a faulty issue that was overlooked by bitmain since the kit wasn't popular.

I have ran S3's oc'd on CX500 for over 1.5 years with zero problems.
legendary
Activity: 1022
Merit: 1003
4 cables are not required on the S3 for any clocking, and it is highly unlikely that would have prevented the damage to the cables.  Also, the S3+ upgrade kits were actually just as if not more stable/reliable than the original S3/S3+'s offered by Bitmain.  Wish they would do more upgrade kits, I loved mine and still run them.
sr. member
Activity: 362
Merit: 250
I use the CX500 on my s3's. It's way cheaper to buy many of them. I ran my s3's at 250 until recently went back to stock to accommodate summer heat. Cables would only get mildly warm. You can get a CX500 for $25 currently after $20 rebate.
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
hmm interesting, some things there i didn't know.  I run them at stock freqs btw.  Im going to put them on the dells just to be safe.  Thanks.

Just remember to un-plug and re-plug the 6 pin PCI-E cables every blue moon ..... or whenever you do your power reboots.


@ op

 this is good advice   especially if it is humid by the miners.
hero member
Activity: 518
Merit: 500
hmm interesting, some things there i didn't know.  I run them at stock freqs btw.  Im going to put them on the dells just to be safe.  Thanks.

Just remember to un-plug and re-plug the 6 pin PCI-E cables every blue moon ..... or whenever you do your power reboots.
full member
Activity: 198
Merit: 102
loose connections  cause  arc and melting. corrosions to the cable can occur over time.


also    if you run at freq 237 or higher the s-3's pull juice at 400 watts.  so 800 watts on a 850 watt psu is hard on the psu . do it for months and the psu cap's age.  So after 6 months you have maybe a 780 watt psu.


I run 2 s-3's on 1000 watt psu.  a seasonic platinum been running about 2 or 3 days less then a year.  24/7/365  no worries.

check this review

http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=126

on page 2 when you go up to  849 watts dc  the efficiency drops like a mofo.  down from 83% to 80%

BTW 83% sucks and that is at 681 watt dc .

 I know 2 s-3s pull more then that  my guess is you pushed it too hard too long.

hmm interesting, some things there i didn't know.  I run them at stock freqs btw.  Im going to put them on the dells just to be safe.  Thanks.

i run at 218 freq btw
legendary
Activity: 4256
Merit: 8551
'The right to privacy matters'
loose connections  cause  arc and melting. corrosions to the cable can occur over time.


also    if you run at freq 237 or higher the s-3's pull juice at 400 watts.  so 800 watts on a 850 watt psu is hard on the psu . do it for months and the psu cap's age.  So after 6 months you have maybe a 780 watt psu.


I run 2 s-3's on 1000 watt psu.  a seasonic platinum been running about 2 or 3 days less then a year.  24/7/365  no worries.

check this review

http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=126

on page 2 when you go up to  849 watts dc  the efficiency drops like a mofo.  down from 83% to 80%

BTW 83% sucks and that is at 681 watt dc .

 I know 2 s-3s pull more then that  my guess is you pushed it too hard too long.
legendary
Activity: 1498
Merit: 1030
Possible the "other" connector on the board with the fried connector got loose.
full member
Activity: 198
Merit: 102
How did you have the power plugged in?  Did you have 2 pcie plugs plugged in from same power cord on each side?   IE one power cord and split into 2 pcie plugs.
The corsair has 4 individual cables with pcie connectors coming out of it, i have two in each miner.  I don't know about the rails on the inside of the PSU though.  They may be basically together inside the case.  Im going to switch to those Dells as soon as I can get some PCIe cables in.

I am thinking that there was a voltage surge and your PSU got damaged and let too much current into the miner. Generally its not a good idea to keep miners running during a storm.
Yes I agree that is probably what happened to the one with the board that quit working.  Its in the server room at my office.  Living in FL, I can't unplug during every storm or it would never be plugged in LOL.

Since all the heat seems to be concentrated at the connection you should of had a bad connection on the pin or corrosion that caused them to over heat.  S3+ should not have problems like that even with 18 AWG.
I pay pretty close attention to detail and I didn't see anything out of the ordinary on the connector or the pins.  I also thought an s3+ shouldn't have problems like that on an 850 watt psu, granted there we're 2 s3s on this psu but still.
legendary
Activity: 1456
Merit: 1000
How did you have the power plugged in?  Did you have 2 pcie plugs plugged in from same power cord on each side?   IE one power cord and split into 2 pcie plugs.
hero member
Activity: 882
Merit: 500
Where am I?
Since all the heat seems to be concentrated at the connection you should of had a bad connection on the pin or corrosion that caused them to over heat.  S3+ should not have problems like that even with 18 AWG.
legendary
Activity: 3808
Merit: 1723
I am thinking that there was a voltage surge and your PSU got damaged and let too much current into the miner. Generally its not a good idea to keep miners running during a storm.
hero member
Activity: 637
Merit: 502
Change your cable for 16 awg. Use a kill-a-watt to monitor if your miner is working properly.
full member
Activity: 198
Merit: 102
I have an S3+ and an S1 converted to S3.  We had a storm and my hash rate dropped so I went to see what was up and I found this....





Power supply is a Corsair TX850 btw.

So the drop in hash rate was the S1->S3 conversion apparently had one of the boards fail most likely from the storm.  The S3+ power cable was way hot when I touched it and I unplugged it and the the connector basically fell apart.  Now the 850 should be plenty to run these two.  What am I doing wrong here?  The burnt ends were the negative cables if that helps.

Now see this pic:


This is the S1/S3, its connectors seemed to get a lil warm, enough to discolor the plastic but they weren't very warm to the touch at the moment but the warm was on the positive side.  Not negative like the other miner.  I don't get it.  

Lastly, I'm going to change to these PSUs with these adapters.


What gauge wires should I be using and how many PCIe connections should i take off each of these Dell PSUs?

Thanks for the help
Jump to: